Timeout for pending state?

I have a OneFlow cluster that scales-up whenever a certain requirement is fulfilled. Since the cloud is quite “full” sometimes there are no available hosts to deploy new VMs and they are stuck in pending state. In this case the requirement above is never fulfilled and the cluster tries to scale-up endlessly, resulting in tens of VMs in pending state.
I would like to know if it is possible to tell ON to automatically cancel VMs if they are not scheduled within a certain period of time.


We don’t have any option to perform that pending VMs “purge”. But you could run a cron script that looks for VMs in the pending state and have an STIME older than a certain threshold.

The script should be easy in bash with the onevm show -x output, or using the ruby/java api: